Title :
An Optimal Orthogonal Expansion for Classification of Patterns
Author :
Nakamura, Yatsuka ; Furuya, Masakazu ; Sunohara, Shoichi
Author_Institution :
Faculty of Engineering, Shinshu University
Abstract :
In this correspondence, the Karhunen-Loève expansion is generalized for multiclass classification and some optimal properties desirable for the feature extraction process are proved. This expansion is based upon a weighted covariance matrix with the interclass and intraclass distances.
Keywords :
Classification of patterns, eigenvectors of covariance matrices, interclass and intraclass distances, multiclass classification, orthogonal expansion of patterns.; Application software; Art; Calculators; Covariance matrix; Digital arithmetic; Equations; Feature extraction; Laboratories; Logic programming; Merging; Classification of patterns, eigenvectors of covariance matrices, interclass and intraclass distances, multiclass classification, orthogonal expansion of patterns.;
Journal_Title :
Computers, IEEE Transactions on
DOI :
10.1109/TC.1977.1674793
